J . Am. Chem. SOC.1981, 103, 6121-6731
6727
Plasticity of the DNA Double Helix' C. K. Mitra, M. H. Sarma, and Ramaswamy H. Sarma* Contribution from the Institute of Biomolecular Stereodynamics. State University of New York at Albany, Albany, New York 12222. Received February 9, 1981
Abstract: In order to determine the detailed dynamic spatial configurationof long dG-dC stretches in double-stranded DNA, magnetic shielding constants were derived for a poly(dG-dC).poly(dG-dC) double helix from the x, y , and z coordinates of A-DNA, alternating B-DNA, Arnott and Hukins' B-DNA, C-DNA, D-DNA, the vertical double helix, the Zl-DNA, the Z2-DNA, and Levitt and Dickerson's propeller twist DNA's taking into account the contribution to shielding from the diamagnetic and paramagnetic components of atomic magnetic anisotropy, as well as the ring-current effects. The computed shielding values for all the forms were examined vis-a-vis the experimentally observed solution nuclear magnetic resonance shift data for poly(dG-dC).poly(dG-dC) in high and low salt solutions. Among the ten different spatial configurations examined, the data indicate that in high salt solution the time-average structure is essentially identical with Z1-DNA, except that local fluctuations about dC is such that the xCNof dC in the solution structure is about 24-25' rather than the 21' in the x, y , and z of Z1-DNA. It is further shown that the time-average solution structure is very close to the real structure with finite lifetime, and it is not the average of some widely different spatial configurations in the fluctuation itinerary. In low salt solution at 81 'C, a temperature 8 OC below the onset of melting, the structure is best described as an equilibrium between high populations of Arnott-Hukins' B-DNA and a model in which the dG may adopt a syn conformation. In the Arnott-Hukins B-DNA, under our conditions, the base pairs could be either flat or mildly propeller twisted. These results along with the recent observations of Patel, Sarma, and their co-workers that in solution the AATT stretches of self-complementaryd-CGCGAATTCGCG assume pronounced propeller twists a la Levitt, Crothers, and Dickerson, the findings of Sarma et al. that changing xCNfrom 80 to 120' causes a right- to left-handed transformation, and the revelation by Rich et al. and Felsenfeld et al. that methylation of the bases promotes B Z transition clearly suggest the rich plasticity in the structure of the DNA double helix and its ability to assume sequence and ionic strength dependent distinct spatial configurations. This vivid demonstration of the plasticity denotes the necessity of paying serious attention to the concept of long-range allosteric transformations in DNA, propounded by Crothers, as a mechanism for the control of genomic expression.
-
The blueprints of the mechanism of life are preserved in genomic DNA. Little do we know about the three-dimensional dynamic solution geometry of any genomic DNA and how the structure controls expression. There have been a plethora of studies on model systems. Fiber diffraction studies in which the atoms are poorly resolved, single crystal crystallographic studies of small DNA fragments, and theoretical studies have led to the suggestion that DNA can exist in diverse spatial configurations, and one does not know how valid are these structures derived from solid state and theoretical studies for polynucleotide double helices in solution. In the present work, we undertake a systematic examination of the solution structure of the double helix poly(dG-dC).poly(dGdC) under high and low salt conditions. Pohl and Jovin2 demonstrated that poly(dG-dC).poly(dG-dC) undergoes a salt-dependent conformational transition. They showed that at low salt concentrations the circular dichroism spectra of poly(dG-dC).poly(dG-dC) and DNA with high dG dC content are similar and the spectra of the synthetic duplex undergo inversion at 4 M NaCI; such inversions can be caused by the addition of mitomycin3 or changing to a solvent of ethanol and water.4 Ethidium has been shown to bind preferentially to the low salt form.s The first qualitative insight about the structure of poly(dG-dC).poly(dG-dC) at high salt concentrations came from the work of Patel et aL6 Their nuclear magnetic resonance ( N M R ) studies suggested that in these structures the symmetry unit repeats every two base pairs, the base pairing to be likely of the Watson-Crick type, and that every other glycosidic torsion angle and phosphodiester linkage adopts values other than those in B-DNA. Extrapolation of the results from single crystal crystallographic studies of hexamer' and tetramer d u p l e ~ e s of *~~
+
For paper 1 in this series, see ref 11. F. M. Pohl and T. M. Jovin, J . mol. Biol., 67, 375 (1972). C. M. Marcado and M. Tomasz, Biochemistry, 16, 2040 (1977). F. M. Pohl, Narure (London), 260, 365 (1976). F. M. Pohl, T. M. Jovin, W. Baehr, and J. J. Holbrook, Proc. Natl. Acad. Sci. U.S.A.,69, 3805 (1972). (6) D. J. Patel, L. L. Canuel, and F. M. Pohl, Proc. Natl. Acad. Sci. U.S.A.,76, 2508 (1979). (7) A. H.-J. Wang, G. J. Quigley, F. J. Kolpak, J. L. Crawford, J. H. van Boom, G. van der Marel, and A. Rich, Nature (London),282, 680 (1979). (1) (2) (3) (4) (5)
0002-7863/8l/l503-6727$01.25/0
dG-dC sequences, as well as data from fiber studies,I0 indicated that the polymer duplex in high salt solution may exist as a left-handed double helix. Recently this laboratory" carried out a very detailed computation of the magnetic shielding constants from the x, y , and z coordinates of the left-handed Z-DNA' and the right-handed Arnott and Hukins' B-DNA (herein after called A / H B-DNA) and compared them with the experimental N M R data for poly(dG-dC).poly(dG-dC) in high salt solution. The computed shielding constants'' for Z-DNA showed a remarkable agreement with the experimental data and that for A / H B-DNA showed significant deviations. This led to the conclusion that poly(dG-dC).poly(dG-dC) in high salt solution can take up the Z configuration and does not assume the structure of A / H BDNA. However it is vital to point out that this does not mean that Z-DNA is the only structure that poly(dG-dC).poly(dG-dC) in high salt solution can assume. In order to reach this conclusion one has to carry out similar studies of other possible models and show that N M R can distinguish among these models and that the experimental data provide a unique fit to Z-DNA. For example, OlsonI2 has advanced a right-handed vertical double helix in which the base planes are almost parallel to the helix axis as a model for the high salt form. Klug et have advocated an alternating B-DNA model for alternating purine-pyrimidine se(8) J. L. Crawford, F. J. Kolpak, A. H.-J. Wang, G. J. Quigley, J. H. van Boom, G.van der Marel, and A. Rich, Proc. Narl. Acad. Sci. U.S.A.,77,4016 (1980). (9) H. Drew, T. Takano, S. Tanaka, K. Itakura, and R. E. Dickerson, Nature (London),286, 567 (1980). (10) S. Arnott, R. Chandrasekaran, D. C. Birdsall, A. G . W. Leslie, and R. L. Ratliff, Nature (London), 283, 743 (1980). ( 1 1) C. K. Mitra, M. H. Sarma, and R. H. Sarma, Biochemistry, 20,2036 (1 98 1). (12) W. K. Olson, Proc. Natl. Acad. Sci. U S A . , 74, 1775 (1977). (13) A. Klug, A. Jack, M. A. Viswamitra, 0. Kennard, Z. Shakked, and T. A. Steita, J . Mol. Biol., 131, 669 (1979). (14) S.Arnott and D. W. L. Hukins, Biochem. Biophys. Res. Commun., 47. 1504 (1972). '(15) S.'Ar&tt, R. Chandrasekaran, D. W. L. Hukins, P. J. C. Smith, and L. Watts, J . Mol. Biol., 88, 523 (1974). (16) D. A. Marvin, M. Spencer, M. F. H. Wilkins, and L. D. Hamilton, J . Mol. Biol., 3, 547 (1961).
0 1981 American Chemical Society
6728 J. Am. Chem. SOC.,Vol. 103, No. 22, 1981
Mitra, Sarma, and Sarma
Figure 1. Stereographic perspective of A-DNA (a) and alternating B-DNA (b) for poly(dG-dC)-poly(dG-dC). Note that in Klug's alternating B model
the base pairs are significantly propeller twisted. The authors1*have never noticed this important nuance in their structure. Table I. Torsion Angles in the Various Modelsa ~~
backbone
g
sugar pucker
furanose
r,
r4
glycosyl torsion
x
37 5
324 334
21 38
anti anti
26 33
42
327
15
12
anti
76
356 356 356 0
25 25 24 340
325 325 325 30
33 33 33 329
342 341 341 20
,E
319 8
36 340
357 24
338 340
223
2E
334
35
330
193
92
'E
345
358
163
146
,E
336
39
ro
r,
,E
4 326
334 17
293
'E
326
214 200 208 178
313 321 298 295
,E ,E ,E 'E
68 190
168 179
294 48
04E-'E
138
56
221
181
93
157
259
147
66
model
W'
@'
A-DNA~ Alt-BDNA' 3'pG5'p Alt-BDNA' 3'pCSp B-DNA~ C-DNA~ D-DNAe ~~son'sf B-DNA Levitt-DNAg ZI-DNA~ 3'pG5'p zI-DNA~ 3'pG5'p 22-DNAh 3'-pG5'p 22-DNAh 3'pC5'
314 278
@
w
46 59
208 151
275 300
,E
143
65
172
155 161 142 198
156 156 156 86
36 37 69 58
273 291
170 256
108 99
80
266
55 74
$'
$
178 192
83 99
227
200
265 254 259 268
a All angles are in degrees. Note that ,E ='E. Reference 17. Reference 8.
Reference 14.
quences; also, the D-DNA is supposed to be true for such sequences. In this paper we attempt a comprehensive and systematic study of the NMR shielding patterns for C G sequences for the diverse spatial configuraitons of D N A whose x , y , and z coordinates are available. From a comparison of these theoretical shielding profiles with the experimental data and introducing proper modifications
Reference 13.
r2
anti anti anti highanti
82 83 83 122
39 7
anti syn
48 248
17
6
anti
21
16
334
25
syn
24 1
322
26
359
anti
33
Reference 16. e Reference 15.
f
Reference 12.
into the starting static structures, we synthesize the dynamic spatial configurations of D N A double helixes in solution. To provide the background and to drive home the geometric differences among these diverse structures in Figures 1 through 5, we stereographically illustrate for poly(dG-dC)-(dG-dC) the structures of A-DNA, alternating B-DNA, A / H B-DNA, CDNA, D-DNA, Olson's B-DNA, Zl/Z2-DNA, and Levitt and
J . Am. Chem. SOC.,Vol. 103, No. 22, 1981 6729
Plasticity of the DNA Double Helix
b. Figure 2. Stereographic perspective of Arnott and Hukins B-DNA (a) and C-DNA (b) for poly(dG-dC).poly(dG-dC).
Dickerson propeller twist models.”J8 In Table I, we have derived the torsion angles from the coordinates, and the nomenclature is explained in Figure 6. The x, y , and z coordinates for deriving the above torsion angles and for deriving the magnetic shielding constants were obtained from Arnott and c o - ~ o r k e r s and ~~,~~ Marvin et a1.16 for A, B, C, and D forms (Figure la, 2, and 3a). For the alternating B model (Figure lb) it was taken from Klug et aLi3 Those for Olson’s B-DNA (Figure 3b) were kindly provided by Olson, and those for Z1- and Z2-DNA (Figure 4) were kindly provided by Rich; those for Levitt (Figure Sa) and Dickerson (Figure 5b) were kindly provided by the authors. The only change we have introduced into some of these original coordinates was to substitute G for A and C for T. For example, the alternating B-DNA was originally proposed by KlugI3 for the poly(dA-dT).poly(dA-dT) sequence. We modified the coordinates by replacing A with G and T with C (Figure lb) mainly to investigate whether poly(dG-dC).poly(dG-dC) can really access the alternating B-DNA structure in solution. In the DickersonI8 model, we confined ourselves to the coordinates from the central G A A T T C stretch of t h e self-complementary dCGCGAATTCGCG double helix. This is because in the Dickerson18 crystal structure it is this domain which displays significant propeller twists. Methodology The chemical shift of a given nuclei, for example, a proton, in the N M R spectrum is strongly influenced by the local geometric and chemical environment of the nuclei. An examination of Figures 1 through 5 clearly illustrates the significant geometric (17) M. Levitt, Proc. Narl. Acad. Sei. U.S.A.,75, 640 (1978). (18) R. Wing, H. Drew, T. Takano, C. Broka, S. Tanaka, K. Itakura, and R. E. Dickerson, Narure (London),287, 755 (1980).
differences among the various helixes. Even though A-, B-, C-, D- and Olson’s B-DNA’s are made up of a monomer repeat unit, the A and Olson’s forms have jE sugar pucker; the sugar pucker of B-, C-, and D-DNA lie in the 2E domain. In the A form, the xCNis low anti and the bases are tilted about 19’ to the helix axis; in the Olson’s form, the xCNis high anti, the bases are almost parallel to the helix axis, and the structure has a vacant inner core of about 35 8, in diameter. The B, C, and D forms belong to a close family of structures, as an examination of the torsion angles in Table I reveals. Even though the alternating B-DNA and Z1 /Z2-DNA’s have a repeat unit of a dinucleotide,the former generates a right-handed helix, but the latter generates left-handed helixes; in the former, the G is anti and in the latter it is syn. Furthermore, there are significant differences in the various torsion angles (Table I) between the alternating form and the Z forms, and these create morphological differences in the double helix (Figures 1b and 4). It should be noted that even though both alternating B- and Z-DNA’s are antiparallel, in the Z forms at each base pair there is local parallelism because the sugar direction undergoes local inversion to accommodate a syn dG to enter into Watson-Crick base pairing in such a way that the glycosyl orientations are cis. A principal feature of the Levitt” model is the propeller twist (Figure 5 ) between the base pairs. In the DickersonI8 model for the self-complementary d-CGCGAATTCGCG this twist can be seen clearly in the middle portion and becomes vanishingly small toward the ends of the helix (vide infra, Figure 9). An important feature of the Dickerson model is that it has no diad axis, and each of the 24 nucleotidyl units has a significantly different spatial configuration, so much so that the two complementary strands are not only nonequivalent but they are structurally revealingly different (Figure 5b).
6730 J . Am. Chem. SOC.,Vol. 103, No. 22, 1981
Mitra, Sarma, and Sarma P
Figure 3. Stereographic perspective of D-DNA (a) and Olson’s vertical double helix (b) for poly(dG-dC)-poly(dG-dC).
We have described elsewhere in extenso the assumptions” and principles involved in the computation of shifts, i.e., magnetic shielding constants, for a double helix. However, to make this presentation as self-contained as possible, we provide a brief summary here. We assume that the chemical shift of the central base-paired nucleotides in a heptamer duplex is the shift of protons dCGCGCGC d-GCGCGCG
in any nucleotide unit in the polymer duplex and that there are no end effects. This assumption is reasonable because the polymer contains close to 100 base pairs and there is translational symmetry; the chemical shifts are not significantly affected by units beyond the third neighbor. We assume that the heptamer duplex can exist in any of the ten spatial configurations illustrated in Figures 1-5. In Figure 7 we have schematically drawn a heptamer segment. The chemical shifts of the central cytidine Co (Figure 7) will be affected by the remaining 13 nucleotide units.” Extensive calculations in this laboratory and that of Pullman and Gie~sner-Prettre’~ have shown that nucleotide units as far away as the sixth neighbor can affect the shifts. However, the chemical shifts are not significantly affected by nucleotide units beyond the third neighbor. The contribution to the chemical shifts originate from (a) ring-current effect of the bases, (b) the diamagnetic component of the atomic magnetic anisotropy of the bases, and (c) the diamagnetic and paramagnetic components of the atomic magnetic anisotropy of the sugar-phosphate backbone. Shielding constants for a given proton of a nucleotide unit in structures like the ones in Figures 1-5 can be computed from x, y , and z coordinates taking (19) B. Pullman and C. Giessner-Prettre, private communication to R.H.S.
all the above contributions into account.’ 1*20-22 The calculated shielding constant essentially provides the magnitude and direction of shielding a proton in a nucleotide unit such as Co (Figure 7) will experience as the unit is moved from an isolated environment to that in an organized structure like the ones in figures 1-5. In such a calculation one cannot include the contribution to shielding from the parent nucleotide unit to which the proton belongs; i.e., Coshould be excluded when computing the effect on Co from the remaining 13 units in configurations Figures 1 through 5 . Go should be excluded when computing the effect of the remaining 13 units on Go. Here one is assuming that the conformation of the isolated mononucleotide is the same as it is in the various organized structures (Figures 1-5). This is not true; hence, a calculation at a second level should be undertaken to correct for the effect of this on shifts, and this is presented later. The ring-current constants and the magnetic anisotropy tensor elements for the calculations were kindly provided by Pullman and Giessner-Prettre before p ~ b l i c a t i o n . ~Using ~ the methods described elsewhere11*2”2and from the x , y , and z coordinates, we have calculated the magnetic shielding constants for CH5, CH6, CHI’, GHS, and GH1’ for the central Co-Go base pair in a heptamer segment (Figure 7) of poly(dG-dC).poly(dG-dC) for the ten different spatial configurations it may display (Figures 1-5). For these computations we have taken into account contributions to shielding from ring-current effects, as well as effects (20) C. K. Mitra, R. H. Sarma, C. Giessner-Prettre, and B. Pullman, Int. J . Quant. Chem., Quantum Biol. Symp., 7 (in press). (21) C. K. Mitra, M. M. Dhingra, and R. H. Sarma, Biopolymers, 19, 1435 (1980). (22) D. M. Cheng, S. S. Danyluk, M. M. Dhingra, F. S. Ezra, M. MacCoss, C. K. Mitra, and R. H. Sarma, Biochemistry, 19, 2491 (1980). (23) F. R. Prado, C. Giessner-Prettre, and B. Pullman, J . Mol. Struct.,
in press.
J . Am. Chem. Soc., Vol. 103, No. 22, 1981
Plasticity of the DNA Double Helix
6731
a.
b.
Figure 4. Stereographic perspective of Z1-DNA (a) and Z2-DNA (b) for poly(dG-dC).poly(dG-dC)
from the diamagnetic and paramagnetic components of the atomic magnetic anisotropy. The results of the computations are presented in Table 11. Before one examines Table 11, we should point out that in the first paper in this series," where we presented for the first time the NMR methodology to handle the details of a double helix, we have presented an extensive table which indicates the magnitudes of the various factors such as ring currents and the paramagnetic and diamagnetic component of the atomic anisotropy from various nucleotide units in a double helix. In Table 11, the data are presented in a considerably reduced format and indicate only the contribution from the various neighbors. Thus, the vertical columns 1 through 7 respectively provide the contributions from (1) the complementary unit, (2) the total contribution from the two nearest neighbors on the same strand, (3) the total contributions from the two nearest neighbors on the complementary strand, (4) the total contribution from the two next nearest neighbors in the same strand, (5) the total contribution from the next nearest neighbors in the complementary strand, (6) the total contribution from the two next next nearest neighbors on the same strand, and (7) the total contribution from the two next next nearest neighbors on the complementary strand. The last column gives the total change in shift a given proton in a mononucleotide will undergo if it is to become part of any of these organized structures. For example, the last column indicates that CH5 will undergo a total shielding of 1.47 ppm if poly(dG-
dC)-poly(dG-dC) were in the A form; out of this 1.47 ppm, -0.098 is contributed by the complementary G, 1.39 is contributed by the nearest neighbors G , and G-,, and so on. How does one experimentally measure the calculated shifts? One has to experimentally obtain the shifts for the poly(dGdC).poly(dG-dC) duplex in the solvent condition of interest and measure those for isolated mononucleotides, Le., the same solvent conditions at extremely low concentration and high temperature. The difference in shifts between the experimental value for the monomers and the double helix should then be compared with those computed for the various spatial configurations. However, there is a problem. As you may recall, in these calculations the shielding effect of 13 neighboring nucleotide units arranged in a particular configuration on the shift of a nucleotide unit was carried out and we neglected, in this first level of calculation, the effect of its own local geometry on its shifts. This would have been alright if the experimentally measured isolated mononucleotides had a geometry identical with what they had in the organized structure. In fact, a plethora of NMR studies from this and other l a b o r a t ~ i r e have s ~ ~ ~shown ~ ~ that monomers, par(24) C. H. Lee, F. s. Ezra, N. S. Kondo, R. H. Sarma, and s. s. Danyluk, Biochemistry, 16, 3627 (1976). (25) F. S. Ezra, C. H. Lee, N. S. Kondo, S. S. Danyluk, and R. H. Sarma, Biochemistry, 17, 1977 (1977).
6132 J . Am. Chem. SOC.,Vol. 103, No. 22, 1981
Mitra, Sarma, and Sarma
'$4
a
Figure 5. Levitt (a) and Dickerson (b) propeller twist models in stereo for poly(dG-dC).poly(dG-dC). P
5'
\" 3
p
-
m
m
